Re: Questionable implementation of IMG ALT attribute as tooltips

Colin F Reynolds (colin@bespin.demon.co.uk)
Wed, 21 Jan 1998 06:12:06 -0500 (EST)


Date: Wed, 21 Jan 1998 06:12:06 -0500 (EST)
Message-ID: <cX0pCPD2lcx0EwDc@the-net-effect.com>
To: www-html@w3.org
From: Colin F Reynolds <colin@bespin.demon.co.uk>
In-Reply-To: <Pine.OSF.3.96.980119135021.17056A-100000@sable.ox.ac.uk>
Subject: Re: Questionable implementation of IMG ALT attribute as tooltips

In article <Pine.OSF.3.96.980119135021.17056A-100000@sable.ox.ac.uk>,
Tim Bagot <timothy.bagot@keble.oxford.ac.uk> writes
>
>On Sun, 18 Jan 1998, Colin F Reynolds wrote:
>
>> My apologies if this has already been discussed here before (I haven't
>> seen it).
>> 
>> Am I, or am I not, correct in assuming that the intention of the ALT
>> attribute of the IMG tag is to provide ALTernative (textual) content in
>> the event that the image to which the tag refers is unavailable?
>
>Or the browser is not displaying images.

To clarify: when I refer to an image being "unavailable", I mean
"unavailable to the user agent", a situation which can arise for a
number of reasons, including, as I have been reminded, if auto-loading
is off, if the image URL is invalid or incorrect, if the image referred
to is missing, or perhaps if that particular connection times out.

>> If this is the case, then isn't the implementation of ALT text as
>> tooltips a design flaw in the user agent, which should be removed at the
>> earliest opportunity (so as to reduce the amount of content in place on
>> the WWW which makes use of this flaw)?
>
>Quite possibly.

Yes... the reason for my asking was to try to gain some insight as to
whether the concensus within the W3C is such that the implementation of
the IMG ALT attribute as tooltips is, to use the vernacular, "BAD"
(Broken As Designed) and whether pressure might therefore be
legitimately brought to bear upon the developers of browsers who misuse
this attribute in this way.

Assuming that it is taken as read that using IMG ALT as tooltips _is_
BAD, which itself needs clarifying, an immediate acknowledgement from
those developers who currently have browsers in the field which feature
this erroneous behaviour, together perhaps with an undertaking that the
fault will be rectified as soon as is feasible, would help to clarify
matters for the authoring community as a whole.

Alan Flavell's treatise on the usage of ALT within HTML <URL:http://ppew
ww.ph.gla.ac.uk/%7Eflavell/alt/alt-text.html> currently bears a
placeholder only under the heading "ALT text as tooltips" (my experience
of Alan tells me that for him _not_ to express an opinion says a lot ;)

There has been some discussion on ciwah about this topic, none of which
has arrived at any satisfactory conclusion.

The W3C could help by providing a lead in this matter. If, of course, it
has already done so, I've not seen the reference and would be grateful
for being enlightened.

>> Has the ALT attribute been hijacked in this way because of an ambiguity
>> in its definition? If so, I move to resolve the ambiguity.
>> 
>> I agree that, since _some_ images are used as hyperlinks, tooltips may
>> be appropriate for those images. If this is deemed a useful enhancement,
>> then in order to cater for this, one backwards-compatible solution which
>> I can see is the simple addition of a TOOLTIP attribute to the IMG
>> element for use in those cases.
>
>Or, use the existing TITLE attribute of the A element.

I agree that rendering an IMG TITLE as tooltips would certainly _appear_
(to me) to be less troublesome than rendering the ALT attribute in this
way (although it is quite possible that this might cause other
difficulties of which I personally am not aware).

The point, surely, is to ensure that the language definition is
sufficiently precise as to reduce the possibility of BAD
implementations.

In any event, Tim, since it is unclear as to whom your advice is
addressed, this does not help to resolve the problem. Advising content
producers to "use the IMG TITLE attribute where tooltips are desired for
the image" is pointless if the browser developers don't intend to
implement it in this way.

>> As things stand, attempting to create hypertext content which caters for
>> a conflicting interpretation of the ALT attribute is, well, it's driving
>> me nuts, for one thing :)
>> -- 

In article <SIMEON.9801191405.A@sf100.essex.ac.uk>, James Green
<jmkgre@essex.ac.uk> writes
>Don't go nuts over it. Use HTML as described by the W3C and you will 
>have no problem.

If only that were true... sadly, it depends upon my "using HTML" as
described by the W3C to author pages for the WWW, (which I do, to the
best of my ability) _together with_ the browser developers doing the
same. You may have noticed that whilst I can control the former, the
latter is somewhat more troublesome ;)
-- 
Colin Reynolds
Please don't ASS*U*ME anything -
 if you do, you'll risk making an ASS out of U and ME :)